mulan.evaluation.loss
Class RankingLoss

java.lang.Object
  mulan.evaluation.loss.RankingLossFunctionBase
      mulan.evaluation.loss.ErrorSetSize
          mulan.evaluation.loss.RankingLoss

All Implemented Interfaces:

Serializable, MultiLabelLossFunction, RankingLossFunction

public class RankingLoss
extends ErrorSetSize

Implementation of the "ranking loss" ranking loss function. It is basically the size of the error set divided by all possible pairs of relevant and irrelevant labels

Version:

2010.11.05

Author:

Grigorios Tsoumakas

See Also:

Serialized Form

Constructor Summary

RankingLoss()

Method Summary

double	computeLoss(int[] ranking, boolean[] groundTruth) Computes the ranking loss function
String	getName() Returns the name of the loss function

Methods inherited from class mulan.evaluation.loss.RankingLossFunctionBase

computeLoss

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

RankingLoss

public RankingLoss()

Method Detail

getName

public String getName()

Description copied from interface: MultiLabelLossFunction

Returns the name of the loss function

Specified by:

getName in interface MultiLabelLossFunction

Overrides:

getName in class ErrorSetSize

Returns:

the name of the loss function

computeLoss

public double computeLoss(int[] ranking,
                          boolean[] groundTruth)

Description copied from interface: RankingLossFunction

Computes the ranking loss function

Specified by:

computeLoss in interface RankingLossFunction

Overrides:

computeLoss in class ErrorSetSize

Parameters:

ranking - the ranking of the learner for an example

groundTruth - the ground truth of the example

Returns:

the value of the loss function